My thoughts on the Text Detective app

Two new apps have recently appeared in the iTunes Store, dedicated specifically for blind people. The first is Fleksy, the second is Text Detective.

There are many audio demos and reviews of Fleksy, but it is a subject we may come back to either in a later blog entry or podcast.

Text Detective allows a visually impaired person to scan paper-based text using the i-device's camera, have it converted to sensible text output using OCR (optical character recognition) technology and have its contents either automatically read by VoiceOver or made available for review.

What follows is not a review of TextDetective, but it is a few thoughts on the app which I could not condense into a tweet. But first, a description of the app from the iTunes Store.

Text Detective finds and reads text in printed documents within seconds using your iPhone’s camera. The detected text can easily be read, edited, copied and pasted into documents, emails or other apps. It even stores all the scanned text in a history, which allows you to refer to it later. Every step is accessible. Text Detective works best with the clean and crisp text you find in documents, and at a distance of about the length of a forearm. You can use it wherever you go, and you don’t need to worry about network coverage – it even works on the plane. The app is optimized for iPhone 4S, but can be used
with iPhone 4 and iPod touch. If you want to use Text Detective on an iPad, please be aware that the camera position is different to the one on the iPhone/iPod, which might require additional training and practice to align the camera lens and align the text.

I have tried a number of such apps on Nokia phones and the iPhone, and I feel this app works just as well as any other. It is important to support the development of this app because it has been developed for us and it is hoped it will be improved over time.

Here are some reasons why you may want to consider purchasing it:

• It is only £1.49 in the UK Store. Compared to some apps, that is inexpensive, although I am not one for begrudging developers of specialist apps for trying to recoup some of the costs involved in both product development and technical support to users.
• It is completely accessible with VoiceOver as you would expect.
• There are a lot of spoken prompts which help a great deal. The way the app works is briefly this. You hold the camera a little distance away from the page you try to scan. There is no need (unlike some other apps) to activate a "Take Picture" Button because the app uses video streaming technology. This means you do not need to disturb focus while you locate the Button and double tap it, a nice feature. TextDetective will then try to repeatedly scan the text. If no results were found, this is spoken by VoiceOver. You then are immediately able to do something about it, such as raising your arm to a greater height to give the app more of an opportunity of scanning the text. If it then subsequently finds text capable of being interpreted, a percentage count of the OCR progress is given. That spoken guidance is invaluable and is reason enough by itself why this app stands out from the rest.
• The other major advantage is that there is plenty of well written help available. The developers (BlindSight) have naturally realised that the majority of people using this app will have never taken a picture before, and they go to great lengths to give as much help as possible concerning how you might go about that. Short of someone showing you how to take a photograph of a page, the help in this app is the next best thing.

In summary then, I think this is an excellent app. But I would respectfully suggest that people consider the quality of recognition results they receive within the context of similar applications of this nature. With even the most expensive of these, the quality of results will only be good if the person taking the photograph is very accurate and it may take several attempts to get satisfactory results from a single page. I think while the concept of being able to sit on a plane or in a restaurant for example to scan a menu is good in theory, the reality is it may not be quite as simple as that.

In terms of reading mail, I feel such apps are useful to get the sense of a document which has arrived for you to read, and will allow you to sort out what may be "junk mail" or something you really need to deal with. But I don't think anything at this point is going to compete in terms of accuracy with a scanner connected to a computer with high quality OCR technology. As long as we keep that in mind when we purchase such apps, our expectations will not be set too high.